Hacker News: roger_

New comment by roger_ in "Show HN: Forge – Guardrails take an 8B model from 53% to 99% on agentic tasks"

roger_ — Wed, 20 May 2026 02:25:50 +0000

Would putting this between a small model and an agent like Hermes improve performance?

New comment by roger_ in "I turned a $80 RK3562 Android tablet into a Debian Linux workstation"

roger_ — Sun, 17 May 2026 21:32:36 +0000

Ha! I spent time also hacking together Armbian on an old A20 TV box.

Claude was definitely helpful the second time around to help with the DTS.

New comment by roger_ in "I turned a $80 RK3562 Android tablet into a Debian Linux workstation"

roger_ — Sun, 17 May 2026 15:02:19 +0000

Looking forward to testing this!

Is full 3D acceleration eventually possible and how's battery live?

New comment by roger_ in "I turned a $80 RK3562 Android tablet into a Debian Linux workstation"

roger_ — Sun, 17 May 2026 14:50:55 +0000

I love how easy AI makes it to hack devices that otherwise wouldn't be worth the time.

New comment by roger_ in "The context window has been shattered: Subquadratic debuts a 12M token window"

roger_ — Sat, 09 May 2026 16:07:14 +0000

Have they published?

New comment by roger_ in "A polynomial autoencoder beats PCA on transformer embeddings"

roger_ — Fri, 08 May 2026 21:51:04 +0000

So this is PCA in kernel space?

New comment by roger_ in "TurboQuant: A first-principles walkthrough"

roger_ — Mon, 27 Apr 2026 13:56:44 +0000

Really good job!

Do you have more of these writeups?

New comment by roger_ in "The electromechanical angle computer inside the B-52 bomber's star tracker"

roger_ — Sat, 18 Apr 2026 17:27:31 +0000

Thanks, I was looking through the article for exactly that. Does it lock on to a configuration of stars?

Really curious how they did this mechanically.

New comment by roger_ in "Show HN: I made a calculator that works over disjoint sets of intervals"

roger_ — Sat, 18 Apr 2026 16:46:10 +0000

I can see valid uses of this but I also feel like a probabilistic calculator would be more useful.

e.g. the result for the 1 / [-1, 2] example doesn’t tell you how likely each value is and it clearly won’t be uniformly distributed (assuming the inputs are).

New comment by roger_ in "Introspective Diffusion Language Models"

roger_ — Tue, 14 Apr 2026 11:30:14 +0000

DFlash immediately came to my mind.

There are several Mac implementations of it that show > 2x faster Qwen3.5 already.

New comment by roger_ in "Show HN: boringBar – a taskbar-style dock replacement for macOS"

roger_ — Mon, 13 Apr 2026 01:39:46 +0000

Looks great but even $40 for a perpetual license with only two years of updates still seems excessive.

Taskbar, Uber, etc. cost less and have unlimited updates. How is this better?

New comment by roger_ in "Understanding the Kalman filter with a simple radar example"

roger_ — Wed, 08 Apr 2026 21:45:21 +0000

Here's my (hopefully) intuitive guide:

1. understand weighted least squares and how you can update an initial estimate (prior mean and variance) with a new measurement and its uncertainty (i.e. inverse variance weighted least squares)

2. this works because the true mean hasn't changed between measurements. What if it did?

3. KF uses a model of how the mean changes to predict what it should be now based on the past, including an inflation factor on the uncertainty since predictions aren't perfect

4. after the prediction, it becomes the same problem as (1) except you use the predicted values as the initial estimate

There are some details about the measurement matrix (when your measurement is a linear combination of the true value -- the state) and the Kalman gain, but these all come from the least squares formulation.

Least squares is the key and you can prove it's optimal under certain assumptions (e.g. Bayesian MMSE).

New comment by roger_ in "Simple self-distillation improves code generation"

roger_ — Sat, 04 Apr 2026 11:58:13 +0000

Skimmed this but don't have an intuitive understanding of why this works and how temperature and truncation factor in.

New comment by roger_ in "Fast and Gorgeous Erosion Filter"

roger_ — Thu, 02 Apr 2026 11:18:53 +0000

I was expecting something about the morphological erosion operator but this was pretty cool.

Some of the techniques here seem to be motivated by physical processes (e.g. rain). I wonder if that could be taken further to derive the whole process?

New comment by roger_ in "Apple Just Lost Me"

roger_ — Wed, 25 Mar 2026 15:18:19 +0000

Criticize gatekeeping all you want, but I feel it’s safer to recommend a Mac or iPhone to an older, non-technical person than the equivalent Windows / Android machine.

And I’m still able to install any app I want with minimal fuss.

New comment by roger_ in "Ubuntu 26.04 Ends 46 Years of Silent sudo Passwords"

roger_ — Sat, 21 Mar 2026 18:05:18 +0000

Why no need to make it the default? I’m all for rethinking legacy decisions.

It helps 99% of the user base and the security risk seems negligible.

New comment by roger_ in "Mamba-3"

roger_ — Sat, 21 Mar 2026 17:58:42 +0000

Can anyone explain why Mamba models start with a continuous time SSM (and discretize) vs discrete time?

I know the step isn’t fixed, also not sure why that’s important. Is that the only reason? There also seems to be a parameterization advantage too with the continuous formulation.

New comment by roger_ in "Attention Residuals"

roger_ — Sat, 21 Mar 2026 14:03:33 +0000

Great idea and seems quite obvious in hindsight.

Is it guaranteed to have the same effect on vanishing gradients though? What if it put weight 1 on a layer that had a tiny gradient?

New comment by roger_ in "LM Studio 0.4"

roger_ — Thu, 29 Jan 2026 13:55:30 +0000

MLX support on Macs was the main reason for me.

New comment by roger_ in "I'm 34. Here's 34 things I wish I knew at 21"

roger_ — Thu, 22 Jan 2026 12:47:15 +0000

Appeal to nature.